A village in northern Tanzania affected by the paralytic disease konzo was compared to a neighbouring unaffected village to investigate possible causes of high cyanide exposure from cassava associated with the disease. Food shortage caused shortening of the established cassava-processing method in both villages. Intensive trade in cassava also contributed to shortening of processing in the affected village. An experiment revealed that established cassava processing by fermenting crushed roots effectively reduced levels of cyanogenic glucosides but significant amounts of yielded cyanohydrins remained in the flour, even after four nights fermentation. Mean (±s.e.m.) cyanohydrin levels were higher in flour from short-processed roots: 57±6 mg cyanide equivalents per kg dry weight, compared to 36±7 when fermented for four nights (P= 0.04). High cyanide intake was supported by high mean (±s.e.m.) urinary thiocyanate: 490±48μmol/l in the affected village and 350±39 in the unaffected one (not significant). Cyanide exposure was aggravated in the affected population by a very low intake of protein resulting in less available sulphur for converting cyanide to thiocyanate, supported by low urinary sulphate. Urinary inorganic sulphate levels were 3802±369μmol/l in the affected village and 7038±855 in the non-affected village (P<0.001).